v0.2.0 -- Fleet Audio Hub, AI DJ Agent, Audio Analysis, Set Recording, Voice Control
New in v0.2.0
Audio Analysis Engine
- Real BPM/key/energy detection via librosa (Krumhansl-Schmuckler)
- Camelot wheel key notation
- Cue point suggestions from track structure
- SQLite database for persistent analysis
AI Autonomous Mix Agent
- \mixx_ai_set(operation='plan')\ -- plans a full set with BPM curve and energy arc
- \mixx_ai_set(operation='perform_transition')\ -- AI chooses and executes transitions
- \mixx_ai_set(operation='suggest_next')\ -- recommends next track based on current deck state
- Uses ctx.sample() for autonomous reasoning; rule-based fallback
DJ Analytics & History
- Play history, transition logging, personal style profiling
- Track suggestions: what haven't you played recently
- Top keys, transitions, and most-played tracks
Set Recording & Replay
- Record every OSC command as JSONL with beat timestamps
- Replay any recorded set through the OSC bridge
- Export sets as JSON for external tools
Fleet Audio Hub
- /api/v1/cockpit/now_playing\ -- aggregated deck status + recording + external sources
- /api/v1/fleet/register\ -- cross-MCP audio source handoff
- /api/v1/fleet/sources\ -- list connected servers
Voice Control
- \POST /api/voice/command\ -- webhook for speech-mcp
- Natural language to OSC command routing
Webapp Updates
- Settings page: SOTA-compliant LLM provider selector (Ollama/LM Studio/vLLM)
- Cockpit: live polling, recording indicator, fleet source badges
- README overhaul with hero, user-friendly structure, mixxxxx video fork docs